Coherent Phonon Anisotropy in Aligned Single-Walled Carbon

Takayoshi Kobayashi , Zhaogang Nie , Bing Xue , Hiromichi Kataura , Youichi Sakakibara , and Yasumitsu Miyata. The Journal of Physical Chemistry C 201...
0 downloads 0 Views 1MB Size
NANO LETTERS

Coherent Phonon Anisotropy in Aligned Single-Walled Carbon Nanotubes

2008 Vol. 8, No. 10 3102-3108

Keiko Kato,†,X Kunie Ishioka,† Masahiro Kitajima,*,†,‡ Jie Tang,§ Riichiro Saito,| and Hrvoje Petek#,% AdVanced Nano Characterization Center, National Institute for Materials Science, Tsukuba, Ibaraki 305-0047, Japan, Department of Applied Physics, School of Applied Science, National Defense Academy of Japan, Yokosuka, Kanagawa, 239-8686, Japan, InnoVate Materials Engineering Laboratory, National Institute for Materials Science, Tsukuba, Ibaraki 305-0047, Japan, Department of Physics, Tohoku UniVersity, Sendai 980-8578, Japan, Department of Physics and Astronomy, UniVersity of Pittsburgh, Pittsburgh, PennsylVania 15260, and Donostia International Physics Center DIPC, P. Manuel de Lardizabal 4, 20080 San Sebastian, Spain Received April 28, 2008; Revised Manuscript Received July 28, 2008

ABSTRACT By time-resolved reflectivity measurements with sub-10 fs laser pulses at 395 nm, the coherent phonons of aligned bundles of single-walled carbon nanotubes are observed for various polarization directions of the pump and probe pulses. In the isotropic reflectivity measurement, we observe the radial breathing modes, G, and even D modes, while in the anisotropic reflectivity mode, only the G mode appears. A complex polarization dependence of the G band phonon amplitude in the isotropic reflectivity is explained by the superposition of G band phonons with different symmetries.

Quasi-one-dimensional carbon nanotubes (CNTs), rolled-up graphene sheets, have attracted much attention because of their unique electronic and structural properties. Quantum confinement effects have significant influence on their physical properties, such as ballistic transport,1 exciton binding,2-4 and Kohn anomaly.5,6 For an isolated singlewalled carbon nanotube (SWCNT) or aligned SWCNT bundle samples, the photoresponse strength in absorption,7 photoluminescence,8,9 photoconductivity,10 and polarized resonance Raman measurements11,12 depend strongly on the polarization of light with respect to the CNT axis corresponding to the strong (weak) optical transition moment for the parallel (perpendicular) polarization alignment.13 The anisotropic optical properties, which arise from the quasione-dimensional nature of CNTs, may lead to important applications of CNTs as optical antennas.14 * Corresponding author. Phone: +81-46-841-3810. Fax: +81-46-8445912. E-mail: [email protected]. † Advanced Nano Characterization Center, National Institute for Materials Science. ‡ National Defense Academy of Japan. § Innovate Materials Engineering Laboratory, National Institute for Materials Science. | Tohoku University. # University of Pittsburgh. % Donostia International Physics Center DIPC. X Present address: NTT Basic Research Laboratories, Nippon Telegraph and Telephone Corporation. 10.1021/nl801200p CCC: $40.75 Published on Web 09/13/2008

 2008 American Chemical Society

Another important optical property of CNTs is a very fast response of the photoexcited carriers. The carrier-scattering processes, such as electron-electron (e-e), electron-phonon (e-ph), or carrier-impurity, which are important for the transport properties of CNTs, have been widely investigated by photoluminescence8,9 and photoconductivity measurements.10 Because of the high mobility of carriers in graphitic materials and the rapid energy relaxation of the photoexcited carriers, CNTs can respond to an outer field even at terahertz frequencies.15 The scattering properties of high-energy photoexcited carriers are determined by emission of optical phonons on time scales less than 100 fs through strong electron- or exciton-phonon interactions. In the case of CNTs, not only the zone-center phonon modes (the G band at around 1580 cm-1 and the radial breathing mode (RBM) at 100-200 cm-1) but also the zone-boundary phonon (G′ band at around 2700 cm-1) are known to be influenced by strong electron-phonon interaction.16 The G′ band, which has a strong Raman response, appears through a doubleresonance Raman process involving two-phonon scattering.17 When one of the phonon scattering processes in the double resonance is substituted by defect-induced elastic scattering, the D band appears at around 1350 cm-1 in defective CNTs and graphite. In the case of the G band, the curvature of a cylindrical CNT splits the degenerate E2g symmetry G band of graphite

into the higher (G+) and lower (G-) frequency modes. In the semiconducting CNTs, the G+ (G-) mode corresponds to longitudinal optic LO (transverse optic, TO) phonon mode, while in the metallic CNTs, the G+ (G-) mode corresponds to TO (LO) phonon mode. This reversal happens because of the softening of the LO phonon mode in metallic CNTs through the interaction with free electrons near the Fermi level (Kohn anomaly5,6). Because of the chirality dependent electron-phonon matrix element for the LO and TO phonon modes, the relative Raman intensity of the G+ and G- bands depends on the nanotube structure.6,18 Moreover, for each G+ or G- band, the periodic boundary condition in the circumferential direction of a CNT defines the angular momentum of vibrational motion along the nanotube axis of 0, 1, and 2, corresponding to the A, E1, and E2 symmetry Raman active phonon modes.19 Thus, we expect six different G band phonon modes in resonance Raman spectra, which can in principle be observed by polarized Raman spectroscopy.19 In fact, in the previous polarized Raman spectra, these phonon modes could be analyzed into separate Lorentzian peaks.12,20-22 For a particular geometry, Raman measurements for aligned nanotubes can selectively detect only E1 or E2 symmetry phonons.19 Because the Raman cross section of the A symmetry G band mode is much stronger than those of the E symmetry modes and the alignment of CNT bundles is not always perfect, the selection of only E1 or E2 symmetry phonons is difficult to attain in polarized Raman spectroscopy. Coherent phonon spectroscopy has the possibility of separating the A and E symmetry phonons by the choice of the excitation and detection geometries.23,24 In general, we can choose the polarization of the excitation light with respect to the axis of the aligned CNTs. Moreover, we can analyze the reflected probe light in a transient reflectivity measurement into components parallel (R//) and perpendicular (R⊥) to the polarization of the pump. The sum (R// + R⊥) of the two signals in such a measurement gives the isotropic response, while the difference (R// - R⊥) gives the anisotropic response. The purely isotropic response could be obtained by recording the reflected light for 45° angle between the pump and the probe polarizations. We have performed measurements on aligned CNTs for various angles between the pump polarization and the alignment axis while detecting either the R// + R⊥ or the R// - R⊥ components. Henceforth, we will refer to the former as the transient reflectivity and the latter as the electro-optic (EO) sampling configurations. In both measurements, the polarized laser excitation with a pulsed duration shorter than the period of phonon oscillation excited coherent phonon oscillations with different symmetries; the phonon oscillations in turn modulate the optical susceptibility and therefore the refractive index.23,25,26 A symmetry phonons induce an isotropic change of the index, while E symmetry phonons induce a birefringent (ellipsoidal) change of the index.25,26 In the transient reflectivity measurement, both the isotropic and the birefringent modulations of the index are detected through the measurement of R// + R⊥. The EO sampling measurement, however, cancels the isotropic index change by the A symmetry phonons while retaining the birefringent Nano Lett., Vol. 8, No. 10, 2008

Figure 1. Polarization dependence of the sample reflectivity. Solid circles are experimental results plotted as a function of the angle (θ) between the laser polarization and the axis of CNTs. The solid line is a fit to the cos(2θ) function.

change by the E symmetry phonons. Such polarization sensitive measurements allow selective coherent phonon spectroscopy of either A or E symmetry phonons. So far, there have been several published reports on coherent phonons in SWCNTs.27,28 By resonant excitation of specific SWCNTs with a tunable laser source with 50 fs pulse, Lim et al. observed the RBM oscillation in SWCNTs samples.28 Gambetta et al., who employed resonant sub-10 fs visible laser excitation, observed both the RBM and the G band mode oscillations.27 They observed the anharmonic coupling between the RBM and the G band modes through the modulation of the G mode frequency synchronously with the slower RBM oscillation. The randomly oriented samples used in the previous experiments, however, prevented the polarization dependence of signals and the symmetry assignment for the G band to be performed. In this communication, we present a transient reflectivity and EO sampling study of the ultrafast dynamics of coherent phonons in aligned bundles of SWCNTs. With laser pulses shorter than the period of C-C stretching mode (21 fs), we are able to observe the coherent oscillations of the RBM, G, and even D modes in the transient reflectivity geometry, while we observe only the G mode in the EO sampling geometry.24,29 By rotating the polarization direction of the pump and probe pulses independently, we investigate the antenna effect, which gives a large optical absorption matrix element for the parallel polarization and the isotropic (anisotropic) electron-phonon interaction for the RBM (G) band phonon. Although our observation does not cover all of the possible geometries and polarization directions, the present data suggest superposition of the G band phonons of different symmetries. Pump-probe time-resolved reflectivity measurements were carried out under ambient conditions with sub-10 fs UV laser pulses at 395 nm. The excitation pulses were produced by frequency doubling of light from a 65 MHz Ti:sapphire laser oscillator.23 The linearly polarized pump and probe pulses were focused by a 50 mm focal-length mirror to a 10 µm spot on the sample respectively with angles of 20° and 5° from the surface normal. The pump power could be varied up to a maximum of 50 mW, while the probe 3103

Figure 2. Time-resolved reflectivity of SWCNTs taken with the (a) transient reflectivity and (b) EO sampling. The inset is the enlarged signal for the oscillatory components. FT spectrum of the time-resolved reflectivity of SWCNTs taken with the (c) transient reflectivity measurement and (d) EO sampling.

power was kept below 3 mW. For the detection of all Ramanactive phonon modes in SWCNTs, the transient reflectivity scheme was chosen, while for the nontotally symmetric (E1 and E2 symmetry) phonon modes, the EO sampling was chosen. In the transient reflectivity measurement, the intensity of the probe beam before and after reflection from the sample was measured with a pair of PIN photodiodes, and the difference between the signal and the reference photocurrents (∆R) was measured as a function of the pump-probe delay. In the EO sampling, the probe beam, which is 45° polarized with respect to the optical plane, is decomposed into polarization components parallel (R//) and perpendicular (R⊥) to the optical plane after the reflection from the sample. Their difference ∆Reo ) R// - R⊥ is measured as a function of the pump-probe delay. The differential current (∆R or ∆Reo) was amplified by a preamplifier with a band-pass filter set to transmit simultaneously the RBM and the G mode signals. The amplified signal was digitized and recorded with a digital oscilloscope. The differential current (∆R or ∆Reo) was normalized to ∆R/R (or ∆Reo/R) where R is the reflectivity without the pump pulse. The sample used in this study consists of SWCNT bundles synthesized by the laser ablation method and purified by reflux in hydrogen peroxide and filtration. A clean hydro3104

philic glass slide was immersed vertically into the SWCNT/ water suspension, and a self-assembled SWCNT film was formed on the slide.30 The direction of SWCNTs was evaluated optically, as well as by transmission electron microscopy and polarized Raman spectroscopy.30 The reflectivity of the probe pulse from aligned SWCNTs shows a polarization dependence that can be used to define the direction of the alignment. The reflectivity of the probe pulse, without pump laser, is plotted in Figure 1 as a function of the angle (θ) between the laser polarization and the axis of SWCNTs. As expected from the previous work,31,32 the polarization dependence of reflection shows a cos(2θ) dependence with a minimum (maximum) when the polarization direction of the laser is parallel (perpendicular) to the axis of SWCNTs. This result is consistent with the antenna effect.13 After excitation of SWCNTs by the pump light, we can observe the ensuing carrier and coherent phonon dynamics through the delay-dependent changes in the amplitude of the reflected probe light. Figure 2a,b shows the experimental results for the time-resolved reflectivity of SWCNTs taken in the transient reflectivity measurement and EO sampling geometries, respectively. In both part a and part b of Figure 2, the transient photoinduced reflectivity consists Nano Lett., Vol. 8, No. 10, 2008

of two components. The first component is the initial, nonoscillatory response due to the excitation and the subsequent relaxation of excited carriers, while the second component is the oscillatory signal due to the coherent lattice vibrations. To extract the oscillation frequencies, the Fourier transform (FT) of the oscillatory ∆R/R (Figure 2c) and ∆Reo/R (Figure 2d) signals are obtained numerically. The FT spectrum obtained by the transient reflectivity measurement (Figure 2c) shows that the coherent phonon response can be separated into three frequency regions, (1) low-frequency (